Application of mixed-effects models for exposure assessment.

Area of Science:

• Occupational Health and Safety
• Statistical Modeling
• Exposure Science

Background:

• Traditional statistical methods like ANOVA have limitations in estimating variance components for repeated occupational exposure measurements.
• Understanding factors influencing both between-worker and within-worker exposure variability is crucial for accurate risk assessment and control.

Purpose of the Study:

• To evaluate the utility of linear mixed-effects models for occupational exposure assessment using existing datasets.
• To compare the performance of mixed models against traditional regression models in identifying exposure determinants.

Main Methods:

• Re-analysis of three datasets from published surveys featuring repeated occupational exposure measurements.
• Application of linear mixed-effects models to estimate between- and within-worker variance components.
• Inclusion of work characteristics as fixed effects to assess their impact on exposure variability.

Main Results:

• Linear mixed-effects models successfully identified work characteristics influencing exposure levels.
• Accounting for work characteristics significantly reduced between-worker variance by 35-80% across datasets.
• Within-worker variability reduction was observed in specific contexts, such as accounting for work activities in pig farming (25%).

Conclusions:

• Linear mixed-effects models are valuable tools for dissecting exposure variability and identifying key influencing factors.
• Identifying determinants of between-worker variability aids in creating more homogeneous exposure groups.
• Understanding within-worker variability drivers supports targeted hazard control and optimized exposure sampling strategies.